Integrating Uncertainty in Coupled Thermo-Hydro-Mechanical Simulations via a Flexible Workflow Orchestrator

Abstract

Accurately modeling subsurface thermo-hydro-mechanical processes is essential for safe and efficient geothermal field development. Acknowledging the model’s many uncertainties is equal in importance for making a sound use of the simulation results. This paper presents a flexible and agnostic workflow that couples Geomechanical Simulations and Thermal-Flow Simulations with varying initial parameters to evaluate the uncertainty in possible field responses. In this example, a synthetic case-study has been made using FLAC3D and TOUGH3 in order to take advantage of the most recent performance developments. Top-cell displacement are then extracted as outputs and compared with observed deformation data similar to what we get with InSAR measurements. To quantify and reduce the uncertainties impact, we then employ an assisted history matching tool driven by Bayesian Optimization progressively constraining an ensemble of models. This leads to digital twin models that are consistent with both the historical field data while acknowledging the subsurface uncertainty, forming a solid basis for better predictions. The workflow is built around a modular, simulation-agnostic orchestrator, allowing for the integration of set ups using different software and different modelling approaches. This flexibility makes it a practical tool for researchers and operators looking to incorporate THM coupling and uncertainty analysis into their field development strategies.